A review of modeling assumptions, uncertainty, and validation limits in the evolution of sub-10 cm orbital debris populations

1 Introduction

The rapid expansion of the space population over the last decade has led to an acceleration in the generation of debris from residual propellant releases and satellite fragmentation. High-profile events such as the Iridium 33/Cosmos 2251 collision, the Starfish Prime nuclear test, and the Fengyun 1C anti-satellite test have been prominent drivers in the creation of debris evolutionary and environmental models. Debris models pioneered by NASA and the European Space Agency (ESA) have become the foundation for debris prediction, removal, and space safety assessments. Currently, the most well-known and widely used models include EVOLVE (Johnson et al., 2001), LEGEND (Liou et al., 2004), MASTER (Bendisch et al., 2004), and ORDEM (Matney et al., 2025), each playing a key role in understanding the near-Earth debris environment and its evolution. Among these debris, objects smaller than 10 cm pose a significant threat due to their sheer number and extreme velocities.

Estimates of the small-debris population vary by source and modeling approach. Surveillance catalogs track on the order of tens of thousands of objects larger than 10 cm (Spivey et al., 2023), while modeled estimates of the 1–10 cm population range from approximately 500,000 in NASA assessments (NASA Orbital Debris Program Office, 2025) to about 1.2 million in the ESA Space Environment Report (European Space Agency, 2025). Regardless of the exact figure, this population remains largely untrackable and is a dominant contributor to collision risk despite its limited measurability.

Debris models are often cited together and provide strong building points for future models. Currently, sub-10 cm debris is only accurately modeled using outdated flux and in-situ datasets as well as sparse observational data. This flux data originates from missions such as the Long Duration Exposure Facility (LDEF) and the EUropean REtrivable CArrier (EURECA) (Bendisch et al., 2004). Data to increase the accuracy of these models is extremely scarce and overall hard to come by. This reliance on aging datasets represents just one of the many significant sources of uncertainty in small debris environmental assessments.

When developing an understanding of how debris is categorized and managed, it is important to recognize the different model types, assumptions, and uncertainties associated with each debris prediction technique. Broadly, debris models can be split into two defining categories as seen in Table 1, each offering distinct advantages and disadvantages depending on the intended application. Further evaluation is important to compare model predictions within the sub-10 cm regime, as in the studies conducted in (Horstmann et al., 2021; Krisko et al., 2015; Liu et al., 2024). The discrepancies highlighted in these studies form the foundation of this review and underscore the challenges associated with small debris modeling, validation, and characterization. Understanding the origins of these prediction results is essential for future debris environmental and evolutionary risk assessments.

TABLE 1. Comparison of evolutionary and environmental debris modeling approaches.

1.1 Modern challenges in Sub-10 cm debris characterization

Within the 21st century, the push for increased satellite coverage has prompted the design of numerous satellite constellations for numerous applications, including but not limited to communication, space sciences, and astronomy (S7 Academies, 2026). Inspecting the satellite presence in low Earth orbit (LEO) shows a dense population of mega-constellations. As of mid-2026, SpaceX operates more than 10,000 active Starlink satellites near 550 km altitude, and the Eutelsat OneWeb constellation comprises roughly 650 satellites near 1200 km (McDowell, 2026; S7 Academies, 2026). This is only a small sample of the mega-constellation network already employed in LEO, and it is projected that up to 100,000 additional satellites could be launched into low Earth orbit within the next decade (Williams et al., 2021), with more conservative agency estimates near 58,000 (U.S. Government Accountability Office, 2022). This vast increase in satellite presence introduces immense strain on current conjunction analysis efforts and elevates the probability of on-orbit collisions. Because collision and fragmentation rates scale with the density of objects in a given orbital shell, continued mega-constellation growth directly increases the rate at which new sub-10 cm debris is produced. Each catastrophic fragmentation releases large numbers of small, untrackable fragments that current models cannot fully resolve, and a sufficiently dense environment raises the risk of a self-sustaining collisional cascade in which debris production outpaces natural orbital decay (Leal Filho et al., 2025; Kessler et al., 2010). This feedback between constellation density and small-debris production is a primary contemporary motivation for the improved sub-10 cm characterization pursued in this review.

The growing importance of debris characterization has also motivated the development of debris modeling capabilities worldwide. In addition to the widely used ESA and NASA models discussed throughout this review, several global efforts will be briefly discussed, including the Space Debris Environment Engineering Model (SDEEM) in China (Pang et al., 2017), the Debris Analysis and Monitoring Architecture to the Geosynchronous Environment (DAMAGE) model in the United Kingdom (Lewis et al., 2001), and the Debris Environment Long-Term Analysis (DELTA) tool by the ESA (Bastida Virgili, 2015).

In conjunction with recently developed breakup models, commercial observation networks have become increasingly prevalent and are important contributors to global space situational awareness efforts. Companies such as LeoLabs operate ground-based phased-array radar networks that provide high revisit rates and accurate state estimation; however, these efforts are reported to be limited to debris larger than 10 cm in size (Parigini et al., 2025).

Recent review efforts have also examined the broader challenges associated with orbital debris monitoring, tracking, and removal. Reviews such as (Dhinakaran et al., 2025; Letizia et al., 2024) have evaluated the complex tasks evaluating current state-of-the-art optical and radar-based tracking systems, while (Dhinakaran et al., 2025) also dives in depth into current debris removal processes. Similarly (Leal Filho et al., 2025), evaluated debris risk from a space sustainability standpoint. This review on debris risk evaluation also stresses the importance of debris mitigation, stressing the dangers of cascading collision events which can lead to a space environment that is no longer usable. These developments reflect the global interest and need in understanding debris generation, evolution, and environmental risk in an increasingly hazardous near-Earth environment.

Over the course of this study, EVOLVE, LEGEND, MASTER, and ORDEM are evaluated along with their respective validation sources to identify key gaps in how sub-10 cm debris is evolved, generated, and tested, while additional debris modeling frameworks, such as SDEEM, DAMAGE, DELTA, and SDPA are discussed to provide a broader view regarding global debris modeling and mitigation efforts. Published cross-model analysis and comparisons are then synthesized to identify where model predictions converge or diverge. By comparing these debris modeling efforts, this review will highlight key areas that generate uncertainty and diminish confidence in small debris propagation.

1.2 Scope and source selection

The objective of this review is to compare the assumptions, uncertainty sources, and validation limits of the principal models used for sub-10 cm orbital debris. This objective calls for interpretive synthesis across a small, tightly interlinked set of foundational models rather than exhaustive enumeration, so a narrative approach is better suited than a systematic survey, which is designed to aggregate large bodies of comparable studies. The literature was therefore selected to serve this comparative aim. Priority was given to the primary sources that define or implement these models, the datasets used to validate them, and the studies that compare them directly. Such sources reside almost entirely in space-agency publications, established peer-reviewed journals, university research groups, and major conference proceedings, and searches within these venues were directed at the governing model papers, their validation datasets, and direct comparisons among them. Because the debris-modeling literature is densely cross-referenced, the initial set was broadened through examination of the references cited within each selected paper, which traces the lineage of model assumptions and validation data back to their foundational sources. A source was retained when it bore directly on the sub-10 cm regime, whether through model formulation and fragmentation physics, in-situ or observational validation data, or cross-model comparison. Studies confined to trackable debris larger than 10 cm, with no consequence for the small-debris regime, fell outside this scope. This focus on the foundational models and their validation data forms the core comparison of this review, while the broader modeling and mitigation efforts are treated where they bear on the sub-10 cm regime.

2 Orbital debris modeling classification

2.1 Evolutionary models

2.1.1 EVOLVE 4.0 and the NASA Standard Breakup Model

EVOLVE 4.0 represents NASA’s implementation of the Standard Breakup Model and serves as the foundation of fragmentation physics for many of the modern debris prediction models (Johnson et al., 2001). As a long-term evolutionary model, EVOLVE 4.0 also holds the capacity to study post-mission disposal of upper stages and individual satellites (Krisko et al., 2001). The NASA Standard Breakup Model was developed to understand the fragmentation physics behind on-orbit collisions and explosions using observations from hypervelocity impact testing, radar measurements, satellite tracking databases, and historical breakup events. Through the statistically derived relationships, the NASA Standard Breakup Model provides a framework utilized by many orbital debris models, whether it be directly or indirectly.

Orbital debris evolutionary models can predict long-term effects on the space debris environment with reasonable accuracy. The NASA evolutionary model, EVOLVE 4.0, accomplishes this through various power-law fragmentation equations and varying principal assumptions. Utilizing these assumptions and equations, EVOLVE 4.0 holds the ability to model satellite explosions and collisions where collisions are further defined as catastrophic ( K i n e t i c E n e r g y ( K E ) ≥ 40 J / g ) or non-catastrophic ( K E < 40 J / g ) (Johnson et al., 2001). Currently, EVOLVE 4.0 categorizes debris smaller than 8 cm using a statistically derived area-to-mass distribution function derived from hyper-velocity satellite impact tests. Such hyper-velocity impact tests originated from the 1986 Delta-180 deliberate low Earth orbit collision (Anz-Meador et al., 1987). Both scenarios have the ability to generate sub-10 cm debris through the use of the described power law function as shown in Equation 1.where M is the mass of the parent satellite in kilograms, L c is the minimal characteristic length, and N ( L c ) is the number of debris generated that are larger than L c. To further refine the model, debris-generation events such as pressurized-tank ruptures, residual-propellant-induced explosions, and spacecraft destroyed through deliberate detonation were investigated (Johnson et al., 2001). These events led to the creation of another power law relationship as shown in Equation 2.where S is a scaling factor and all other terms are consistent with Equation 1. This relationship was validated for debris greater than 1 cm using the Haystack radar, which detects at altitudes of 600–1100 km. Following fragmentation generation, EVOLVE 4.0 has provided significant improvements on past model revisions by assigning physical properties to individual debris objects through statistical calculation of area-to-mass ratio ( A / M ) and the change in velocity after ejection ( Δ V ) (Reynolds et al., 1999). These parameters have proven to be critical for the long-term propagation of small debris, owing to their strong dependence on atmospheric drag and solar radiation pressure (SRP) (Johnson et al., 2001). The aforementioned statistical calculations were again derived from hyper-velocity impact testing and orbital decay rate analysis, including the 1986 Delta-180 deliberate low Earth orbit collision.

N L c = 0.1 M 0.75 L c − 1.71 (1)

N L c = S 6 L c − 1.6 (2)

Due to the reliance on both explosion and collision fragmentation models in a derived power-law equation, most of the calculated debris falls into the small-size regime. As the parent characteristic length begins to decrease, the resulting number of debris that falls in the sub-10 cm category begins to rapidly increase. While this behavior is physically consistent with fragmentation theory, the underlying power-law relationships are largely derived from outdated collision and explosion data, leading to extreme uncertainties in the sub-10 cm regime.

Despite these limitations, the EVOLVE 4.0 model provides a physically consistent framework that can be used for the accurate modeling and long-term propagation of debris post-fragmentation. By incorporating physical properties such as A / m and Δ V, the EVOLVE 4.0 model can also accurately predict decay rates, which is useful for not only conjunction analysis but also environmental stability studies, as shown in (Jain and Hastings, 2024).

2.1.2 Legend

Building on past efforts to increase the fidelity of debris evolutionary models, the NASA Orbital Debris Program Office began the development of the LEO-to-GEO Environment Debris Model (LEGEND) to provide enhanced long-term propagation of debris across multiple orbital regimes. LEGEND expands upon previous evolutionary models by incorporating updated fragmentation and propagation physics. LEGEND simulates large debris populations over extended time periods using parent object state vectors and, if unavailable, Monte Carlo methods. These Monte Carlo methods allow for the generation of debris populations while maintaining physically consistent orbital distributions (Liou et al., 2004).

LEGEND can be broadly divided into two overarching categories, differentiated by propagation features: GEOPROP and PROP3D. GEOPROP governs the propagation of GEO objects categorized by eccentricity ( e < 1 ), inclination in degrees ( i < 15 ° ), and orbital period in days ( 0.95 < T < 1.05 ). PROP3D, on the contrary, is capable of propagating debris populations across all other orbital regimes, LEO, Medium Earth Orbit (MEO), and Highly Elliptical Orbits (HEO). Both propagators account for the most significant orbital perturbations, including atmospheric drag, lunar gravitational effects, solar radiation pressure (SRP), Earth’s shadow effects, and zonal harmonics to an accuracy of J 4 and ( J 2 ) 2 (Krisko and Liou, 2003). Capturing these dynamic perturbations enhances the LEGEND model’s high-fidelity predictive capabilities.

LEGEND is further broken down into five major loops for historical propagation before 2001. Each loop is a designated pathway for the software to take depending on debris parent origin (i.e., existing intact objects, existing fragments, new breakup fragments, non-breakup fragments (residual propellant debris), and newly launched intact payloads) (Liou et al., 2004). Similar to EVOLVE 4.0, LEGEND relies on the same power law equations (Equations Equation 1 and Equation 2) to define the number of debris created by an explosion/collision. Consequently, LEGEND utilizes the same data sources that these equations were derived from and statistically assigns properties such as A / M and Δ V as described in EVOLVE 4.0. Additionally, these equations retain drawbacks arising from the restricted sub-10 cm debris cataloging due to tracking gaps and limited fragmentation datasets (Johnson et al., 2001).

2.2 Environmental models

2.2.1 MASTER

Further categorization of debris clusters that fall within the 100 μ m size regime is also critical for assessing spacecraft risk and safety. Environmental models such as the Meteoroid and Space Debris Terrestrial Reference Model (MASTER) fill this gap through the utilization of debris flux predictions (Braun et al., 2021). Unlike the evolutionary debris models discussed previously, MASTER does not explicitly simulate or track individual Resident Space Objects (RSOs). Instead, the model reconstructs the debris environment using a combination of datasets, experimental testing, and known fragmentation scenarios.

MASTER represents the debris environment through a series of debris source populations (i.e., paint flakes, ejecta, West Ford Needles, NaK droplets, slag, and Two Line Element Sets (TLEs)) (Bendisch et al., 2004).

One representative debris population included in MASTER is the West Ford Needles. Project West Ford was an attempt by the MIT Lincoln Laboratory in 1962 to improve radio communications systems by deploying a belt of orbiting dipole antennas composed of small copper needles (Lovell and Ryle, 1962). These dipoles were designed to resonate near 8 GHz, allowing ground stations to reflect radio signals from the artificial antenna belt. Approximately 4.8 × 1 0 8 dipoles were deployed near 3600 km altitude in 1963 (Shapiro et al., 1964), though a post-mission analysis later concluded that only 25 to 45 percent dispersed as intended (Liou and Shoots, 2013). With the initial orbit of the West Ford Needles known, this source provides an accurate way to analyze how the debris environment at 1 mm evolves over time using radar detections. By including multiple debris source populations, MASTER developed the capability to reconstruct the orbital debris environment extending from micrometer-scale particles to trackable RSOs. Spatial density distributions, as shown in Figure 1, are generated from datasets obtained from both in-situ and remote-sensing observations (e.g., LDEF, EURECA). These datasets provide important calibration data for debris populations below traditional cataloging thresholds (Braun et al., 2021).

Despite the inclusion of these experimentally derived datasets, the uncertainty in the sub-10 cm region remains prominent. Additionally, small debris populations are highly sensitive to their material properties and fragmentation energies (i.e., ejection velocity, A / M, etc.), which are very difficult to replicate through testing alone. As a result, MASTER relies on datasets similar to those described in EVOLVE 4.0 and LEGEND, which use derived scaling relationships, introducing additional uncertainty into the model.

2.2.2 Ordem 4.0

Following the development of space environmental models such as MASTER, NASA’s Orbital Debris Program Office (ODPO) created the Orbital Debris Engineering Model (ORDEM) series to statistically characterize the near-Earth debris environment. ORDEM utilized in-situ, radar, and optical measurement datasets to estimate debris flux as a function of particle size, velocity, direction, and material density (Kennedy et al., 2022). Over several iterations of ORDEM, from ORDEM96 to ORDEM 4.0, the ODPO has refined ORDEM solutions and modeling techniques to improve the accuracy of environmental reconstruction across multiple orbital regimes. Using these computations, ORDEM can then predict environmental flux for long durations, which is particularly useful for impact risk assessment, shielding design, and overall safety predictions.

ORDEM was originally developed in the 1980s to provide a validated model of the current debris environment. Modern iterations of ORDEM, including ORDEM 4.0, incorporate measurement data obtained from the U.S. Space Surveillance Network (SSN), the Haystack Ultrawideband Satellite Imaging Radar (HUSIR), the Michigan Orbital DEbris Survey Telescope (MODEST), and in-situ measurement data obtained from shuttle orbiter impacts. Using these datasets, ORDEM reconstructs debris populations spanning particle sizes from approximately 10μ m to 1 m, effectively bridging the gap between micrometeorites and trackable RSOs (10 cm) (Matney et al., 2025; Cowardin, 2022).

Recent updates in ORDEM 4.0 incorporate improvements to the density families derived from hypervelocity impacts observed on Space Shuttle surfaces. Such impacts yielded insights into the materials orbiting the Earth and the influence of material properties on impact damage. Despite these improvements, uncertainty remains greatest in the sub-10 cm regime due to limited observational data and reliance on data reconstruction. These limitations have been acknowledged in (Matney et al., 2025), which notes that limited observational coverage and the lack of recent in-situ measurement data limit the accuracy of the model, often requiring reliance on updated analysis of legacy datasets.

2.3 Additional debris modeling frameworks

While EVOLVE, LEGEND, MASTER, and ORDEM remain among the most widely used and popular debris modeling techniques, it is also important to identify several additional international efforts to characterize the debris environment and overall population.

2.3.1 SDEEM model

The SDEEM developed by the Hypervelocity Impact Research Center at the Harbin Institute of Technology in China aims to model the debris environment from 200 to 2000 km from a time period of 1959–2050 (Pang et al., 2017). Similar to MASTER, SDEEM models solid rocket motor slag, NaK droplets, fragmentation debris, and ejecta through the analysis of numerous fragmentation events, solid rocket motor firings, and releases of NaK coolant from nuclear reactors operated in space. It is most noticeable that through model outputs, SDEEM was able to compare relatively well to other established environmental models, such as MASTER and ORDEM, for debris sizes greater than 10 cm. However, when modeling the smaller debris population SDEEM, ORDEM, and MASTER struggled to converge to similar values. This is seen when modeling debris greater than 1 mm ORDEM predicts approximately one order of magnitude greater than both SDEEM and MASTER (Pang et al., 2017). As smaller debris particles are modeled, cross-model divergence becomes increasingly pronounced and none of the models agree on how spatial density is distributed throughout the orbital environment.

2.3.2 DAMAGE model

The DAMAGE debris modeling tool was developed by the University of Southampton to enable the higher-fidelity characterization of the debris environment in higher Earth orbits (Lee et al., 2024). This model represents debris down to 0.1 mm in size at altitudes of 2000 km to GEO, while the main applications are risk assessments for spacecraft operating in higher Earth orbits, including analysis of mitigation measures and spacecraft disposal strategies (Lewis et al., 2001). DAMAGE operates as a long-term evolutionary debris modeling framework by combining historical launch records, future traffic predictions, collision-risk assessment tools, and mitigation strategies to evaluate the long-term stability of the higher Earth orbit environment.

2.3.3 DELTA model

Developed by the ESA, the DELTA tool was designed to analyze the long-term evolution of the space debris environment in LEO, GEO, and MEO over various time intervals (Bastida Virgili, 2015). At the core DELTA utilizes a fragmentation foundation of the NASA Standard Breakup Model and initial debris populations taken from the MASTER-2009 model to evaluate the debris environment. As a result of inherited properties, DELTA retains the debris size modeling limitation of MASTER-2009, restricting debris characterization to particles greater than 1 mm while modeling fragmentation events, NaK droplets, and solid rocket motor slag (Lee et al., 2024). Through these capabilities, DELTA provides its most useful output of collision prediction. As described from (Bastida Virgili, 2015), the collision prediction tool can be used for improved active debris removal (ADR) selection and showed removal of dangerous debris reduced long-term debris population growth.

2.3.4 SDPA model

The principal Russian effort is the Space Debris Prediction and Analysis (SDPA) model, developed by Nazarenko using a statistical approach grounded in the theory of satellite ensemble motion, in which the debris population is treated as a continuous medium rather than as individually propagated objects (Nazarenko, 1997). The openly documented engineering implementation, SDPA-E, provides fast estimation of spatial density, cross-sectional flux, collision-velocity distributions, and flux prediction across the LEO and GEO regimes (Nazarenko and Menshikov, 2001). However, SDPA-E considers only objects larger than approximately 1 mm, so like DELTA it does not extend into the deep sub-millimeter regime, and its small-debris treatment is statistical and far less documented in accessible, peer-reviewed English-language sources than ORDEM, MASTER, or SDEEM. No modern, openly available English-language Russian sub-10 cm engineering model exists for direct cross-comparison, which itself represents a gap in the international picture of small-debris characterization.

3 Uncertainty sources and validation limitations

3.1 Model driven uncertainty

This section separates uncertainty that originates in the structure and assumptions of the models from uncertainty that originates in the measurements used to calibrate and validate them. Model-driven uncertainty refers to the former, which includes the chosen fragmentation formulation, the functional form of the power-law relationships, and the assumptions required to propagate fragments through the orbital environment. As briefly mentioned in evolutionary models such as EVOLVE 4.0 and LEGEND, current models rely on empirical relationships to characterize breakup events. These relationships include power-law equations that produce physically consistent results. However, when these relationships are extrapolated into the sub-10 cm regime, significant uncertainty arises, because the assumed power-law form is applied well beyond the size range over which it was fitted. The scarcity of measurements that would otherwise constrain this extrapolation is a distinct, data-driven source treated in Section 3.2. The fragment population produced by the break-up model concentrates at low area-to-mass ratios, as shown in Figure 2. Because this distribution is steep, even small variations in the initial break-up parameters can lead to large discrepancies in the predicted debris population (Kessler et al., 2010).

These variations can also directly affect orbital lifetime predictions, because A / M (area-to-mass ratio) is directly proportional to both atmospheric drag and solar radiation pressure, a drag ∝ A / M and a SRP ∝ A / M, as derived from (Vallado, 2013). Therefore, any change in fragment size or number can affect debris orbital decay rates by altering A / M.

These uncertainties are then inherently propagated forward into ejection velocity and A/M definitions. Variability in these two parameters can directly impact the orbital parameters of the characterized debris rings, as seen in (Kirk and Kilic, 2026), thereby affecting predicted debris occupancy regions, particularly in LEO, where drag effects dominate orbital decay. Due to the limited observability of sub-10 cm debris as stated in (Oikonomidou et al., 2025), fragmentation models serve as the primary means of bridging the gap between flux-based millimeter regime models and observable debris greater than 10 cm. This importance and reliance on these models further emphasize the need to characterize model uncertainties and limits.

3.2 Data driven uncertainty

Data-driven uncertainty, by contrast, originates not in the model formulation but in the measurements available to constrain it. This category includes the limited altitude and time coverage, the age, and the sampling bias of the datasets used for calibration and validation. Environmental debris models such as MASTER and ORDEM rely heavily on derived datasets to reconstruct the debris environment from the millimeter-size regime to trackable RSOs (greater than 10 cm). While these models provide strong agreement with physical trends, uncertainty in the sub-10 cm regime still remains prominent due to the limited availability of modern in-situ measurements (Matney et al., 2025). Much of the existing data used within environmental models originates from legacy datasets such as the LDEF, European Retrievable Carrier (EURECA), and Michigan Orbital Debris Survey Telescope (MODEST) (Matney et al., 2025; Bendisch et al., 2004). These datasets, obtained over limited altitude ranges and time periods, produced a sparse sampling of the debris environment (Seitzer et al., 2004). As a result, environmental models rely on extrapolating debris flux distributions beyond validated regimes, introducing uncertainty into environmental predictions (Matney et al., 2025).

The sub-10 cm debris population remains particularly difficult to understand and map due to the sensitivity of the debris size to optical and radar tracking systems. Objects below the 10 cm threshold require indirect estimation methods, such as data extrapolation to reconstruct the debris environment. While recent efforts have attempted to eliminate this discord in data, these datasets remain localized and can not fully capture the debris environment (Kuitunen et al., 2001; Drolshagen et al., 2001).

Despite these limitations, environmental models such as MASTER and ORDEM remain critical tools for spacecraft safety and space environment assessments. The ability of environmental models to span multiple size regimes is invaluable for mission success in the near-Earth environment. Continued improvements in in-situ measurement technologies, optical sensing, and radar detection techniques will further enhance confidence in long-term debris environmental predictions.

3.3 Uncertainty quantification approaches

Uncertainty in orbital debris modeling can be broadly divided into two categories: epistemic and aleatory. Aleatory uncertainties arise from the unknown of future collisions and other fragmentation events. Epistemic uncertainty arises from incomplete knowledge of fragmentation physics, observational data depth, and model formulation approaches, which is discussed in detail throughout this review.

The most common way to quantify uncertainty within debris models is the utilization of Monte Carlo analysis and simulation. A Monte Carlo approach to debris generation and evaluation allows for probability distributions through probability density functions (PDFs) and defined initial parameters (Jang et al., 2024; Liou, 2008). For example, the LEGEND evolutionary model utilizes this Monte Carlo approach to determine parent object state vectors if they are unknown (Liou et al., 2004). Each Monte Carlo output represents one possible result of a collision event. By repeatedly sampling uncertain inputs (the parent state vectors), a distribution of possible outcomes is produced, and an estimation of how input uncertainty affects the debris population can be formulated. Although Monte Carlo analysis helps to quantify aleatory uncertainties, epistemic uncertainties remain large due to biased and limited observational datasets for debris that are below 10 cm. Differences in how debris models are validated and formulated can in turn affect model output accuracy and cross-model convergence. In practice, this input uncertainty is propagated to model outputs by repetition of the simulation across sampled parameter sets and examination of the resulting spread in predicted flux and population, so that the output distribution reflects the assumed input distributions.

3.4 Cross-model divergence in the sub-10 cm regime

Direct comparisons between environmental models reveal substantial quantitative disagreement in the sub-10 cm regime. Horstmann et al. (2021) compared MASTER-8 and ORDEM 3.1 flux predictions across three reference orbits at a 2016 epoch and found that MASTER-8 predicted approximately 2.3 times higher flux than ORDEM 3.1 at 1 cm for an 800 km sun-synchronous orbit, and approximately 3 times higher for a geosynchronous transfer orbit. Below approximately 2 mm, the relationship reverses: ORDEM 3.1 exceeded MASTER-8 by up to 2.5 orders of magnitude at sub-millimeter sizes, depending on orbit. An earlier comparison by Krisko et al. (2015) between ORDEM 3.0 and MASTER-2009 identified the same crossover behavior near 1 cm, with ORDEM flux overtaking MASTER as debris size decreased in all four test orbits examined. Liu et al. (2024) independently reproduced this pattern with a third model (SDEEM2019), confirming that the divergence is systematic rather than an artifact of any single implementation; the authors attributed the differences to distinct modeling approaches for uncatalogued debris. Critically, both MASTER and ORDEM employ modified versions of the same underlying fragmentation model (the NASA Standard Satellite Breakup Model), so the disagreement does not originate from breakup physics but from differences in source population treatment, calibration datasets, and scaling methodologies. A study by (Pang et al., 2017) also showed similar results with large divergence in the smaller debris populations. To provide a consolidated view of the discrepancies seen through numerous size and altitude regimes, Table 2 summarizes cross-model comparisons findings shown in (Pang et al., 2017) for the sub-10 cm debris population for the environmental models discussed beforehand. Absolute flux and population counts depend on the specific orbit, epoch, and configuration of each model run and are not uniformly reported across the source studies; Table 2 therefore synthesizes the published relative divergence between models, which is the configuration-independent quantity most relevant for the assessment of model reliability.

TABLE 2. Cross-model comparison of orbital debris spatial-density predictions across multiple size regimes, based on the SDEEM-2015, ORDEM2000, and MASTER-2009 distributions reported graphically by Pang et al. (2017). Divergence values are estimated by the present authors from the spatial-density-versus-altitude figures in that study, read at the altitudes of maximum separation; the only divergence stated numerically in the source text is the order-of-magnitude excess of ORDEM2000 over the other two models at sizes larger than 1 mm.

It is important to note that Table 2 reflects the older ORDEM2000 and MASTER-2009 distributions reported by Pang et al. (2017), in which SDEEM-2015 is the high model near the 1 cm peak. This is distinct from the recent MASTER-8 versus ORDEM 3.1 comparison of Horstmann et al. (2021), in which MASTER-8 exceeds ORDEM 3.1 by a factor of two to three at 1 cm. The two results are consistent rather than contradictory: they compare different model pairs at different versions and epochs, and the headline factor of two to three quoted throughout this review refers specifically to the current-version MASTER-8 and ORDEM 3.1 comparison. On the evolutionary model side, the IADC Working Group 2 comparison study (Liou et al., 2013) ran six independent models (SDM, DELTA, KSCPROP, LEODEEM, LEGEND, and DAMAGE) from ASI, ESA, ISRO, JAXA, NASA, and UKSA on a common initial population and found consistent qualitative agreement for objects 10 cm and larger, with an average population increase of approximately 30% over 200 years. However, no equivalent multi-model benchmark exists for the 1–10 cm regime. The absence of such a benchmark, combined with the factor of two–three systematic gap between MASTER and ORDEM at 1 cm, underscores that current debris models cannot converge on sub-10 cm population estimates and that mission risk assessments remain sensitive to model selection.

3.5 Model validation limitations

3.5.1 In-Situ validation methods

3.5.1.1 Long Duration Exposure Facility

The LDEF represents one of the most widely utilized validation datasets for environmental debris models. The LDEF was deployed by NASA on 7 April 1984, at an altitude of 475.964 km and an inclination of 28.5° (Stuckey, 1993). The spacecraft provided large in-situ datasets obtained from pits on the surface of the LDEF from hypervelocity debris impacts. These impacts allowed for the characterization of material properties and debris sizes, and still remain one of the most complex datasets for small debris characterization today.

Upon the retrieval and de-orbit of the LDEF onboard the space shuttle, it was expected that most of the impacts from debris would be on the leading edge of the spacecraft, while the trailing edge would hold mostly micrometeorite impacts. However, after further analysis, it was concluded that more than 30% of the impacts on the trailing edge of the LDEF were generated by debris collision, implying that there is a large population of debris in Highly Elliptical Orbits (HEO) (Stuckey, 1993). This is just one of the many important findings the LDEF discovered during its missions to classify the near-Earth debris environment.

Despite its importance and groundbreaking work, the LDEF datasets introduce several limitations when used for debris model validation. As mentioned before, this mission operated at approximately 476 km altitude, limiting its applicability to other orbital regimes, such as Medium Earth Orbit (MEO) and Geosynchronous Earth Orbit (GEO). Second, the mission’s 5.7-year duration provides only a brief snapshot of the debris environment. Since the conclusion of the LDEF experiment, the debris environment has continued to increase exponentially, as shown in Figure 3, driven by the surge in rocket launches and satellite deployments in recent years.

While the LDEF remains one of the most valuable datasets for small debris classification, its lack of spatial sampling introduces uncertainty that is inherently propagated into environmental models.

3.5.1.2 European Retrievable Carrier

Similar to the LDEF, EURECA provided additional in-situ measurements for the small debris environment through the analysis of impact craters. On 31 July 1992, EURECA was launched aboard the Space Shuttle Atlantis and recovered on 1 July 1993, after approximately 1 year in LEO. Like the LDEF, EURECA was also stationed at a fixed altitude to collect debris and micrometeorite flux data, enabling collection of micrometeorite and debris flux data over extended exposure periods (Aceti et al., 1993).

Post-recovery investigations focused on material degradation, spacecraft anomalies, and technology validation studies. Of particular interest was the analysis of impact features on spacecraft surface materials (Aceti et al., 1993). The initial post-flight survey of the spacecraft main body recorded 75 confirmed impact holes larger than 200μ m in the multi-layer insulation (MLI), along with impacts on the scuff plates (Dover et al., 1995). A subsequent detailed optical survey refined this to 71 MLI impact sites with hole diameters from 100μ m to 1 mm, and found that the largest single feature was a 2 mm-diameter hole through a painted-aluminum sign plate, likely caused by a particle in the 1 mm size range; none of the MLI impacts penetrated all 21 layers of the most common blanket type (Drolshagen et al., 1996). These measured impacts were directly used to provide accurate knowledge of the near-Earth environment and provide a basis for environmental particle flux predictions.

The main source of comparison for this mission was ESABASE. ESABASE is a European Space Agency software suite that provides more than 10 applications for space environmental analysis, including ORDEM and MASTER (Bunte et al., 2004). Results from ESABASE and EURECA were compared, showing overall agreement; however, at larger sizes (100–1000 μ m), the predicted and measured results deviated, with the model over-predicting the estimated flux. Additionally, EURECA was only able to measure particle flux for a small size window (100–1000 μ m), restricting validation of debris flux models outside of this range (Aceti and Drolshagen, 1995). As a result, extending debris flux predictions beyond the measured regime requires extrapolation, which introduces significant uncertainty into environmental model outputs (Aceti and Drolshagen, 1995).

3.5.1.3 Modern in-Situ measurement missions

A renewed emphasis on in-situ data collection has emerged. Recent validation missions such as ADLER-1, the International Space Station Space Debris Sensor (ISS SDS), and DEBIE-2 have become the cornerstone for cross-model validation and performance evaluation (Grömer et al., 2025; Anz-Meador et al., 2019; Hamilton et al., 2017; Menicucci et al., 2013). Currently, the small object population modeled in the MASTER-8 model is only validated through the LDEF, EURECA, and the Hubble Space Telescope (HST) mission, which provides limited validation value as the debris environment evolves over time (Grömer et al., 2025).

The ADLER-1 Mission represents one of the most recent data acquisition missions for smaller debris populations. Operating at an altitude of approximately 480 km for 512 days, ADLER-1 was able to characterize small debris populations typically unseen by most ground-based observation systems (Grömer et al., 2025). During a data collection period of approximately 400 days, ADLER-1 recorded 117 impact events, which were able to be compared against the ESA MASTER model for updated validation (Grömer et al., 2025). In comparison, at an impact orientation of less than 20°, the MASTER model predicted five space debris impacts and 18 meteoroid impacts, totaling only 23 total impacts. However, data obtained by ADLER-1 showed a total of 62 detected impacts (Grömer et al., 2025). This discrepancy is not consistent and is highly sensitive to impact orientation. As shown by an impact orientation of 60–80°, MASTER predicted a total of 26 impacts, while ADLER-1 realized only five of these impacts (Grömer et al., 2025). These results demonstrate that the measured debris flux can vary substantially from model predictions depending on the local impact orientation. While ADLER-1 provides valuable insight into the local debris conditions, additional in-situ measurements, such as the ISS SDS mission, have been developed to help provide continuous monitoring of the small debris environment in the ISS orbital regime.

Unlike shorter-duration missions such as the LDEF and EURECA missions, the ISS SDS was designed to provide continuous real-time characterization of the small debris environment (Anz-Meador et al., 2019). Attached to the ESA Columbus module the ISS SDS mission represents the first in-flight demonstration of the Debris Resistive/ Acoustic Grid Orbital NASA-Navy Sensor (DRAGONS) (Anz-Meador et al., 2019).

The goal of this mission was to obtain micro-meteoroid and debris flux measurements within the 50-μm to 500-μm size regime over the course of its minimum of 2-year operational timeline (Hamilton et al., 2017). The SDS was strategically placed so that the normal surface of the sensor will continuously be within 5 degrees of the ram direction for optimal detection (Hamilton et al., 2017). The continuous evaluation of the debris environment that the SDS provides is a valuable source of validation data for environmental models. However, despite these advantages, as mentioned in (Grömer et al., 2025), such efforts are limited to short time intervals and small altitude-inclination regions, limiting the applicability of the datasets. While in-situ data sets provide valuable direct detections within the debris environment, their orbital coverage limits their ability to fully capture the full debris population. To complement these space-based measurements, several ground-based observation facilities have been developed to provide detection along wider orbital regimes.

3.5.2 Ground-based validation methods

3.5.2.1 MODEST

MODEST is an essential data-collection telescope developed by the University of Michigan to characterize the GEO environment. MODEST uses a 2048x2048 charged-coupled device (CCD) with a 5-s integration time to detect RSOs in GEO down to 20 cm in size (Abercromby et al., 2011). For this experiment, objects in GEO were defined as objects having an inclination of less than or equal to 15° (Abercromby et al., 2011). To accurately categorize the RSOs, inertial stares were taken to generate ”streaking” effects created by the relative motion of the debris (Abercromby et al., 2011). These streaks are then converted to right ascension (RA) and declination (DEC) measurements for initial orbit determination (IOD).

Despite these capabilities, several limitations arise when using ground-based observation for small debris detection. As described in (Abercromby et al., 2011), MODEST is limited to Line of Sight (LOS) geometries that lie closest to the anti-solar point while remaining outside of the Earth’s shadow. The anti-solar point (ASP) is defined as the point on the celestial sphere that is directly opposite from the sun (180°) from the direction of the observer. The ASP is selected for observation because of its capability to maximize signal-to-noise ratio while minimizing image saturation from the sun. Another limitation is the short orbital arc provided by the RSOs as captured from MODEST. To derive orbital position and velocity data, methods such as Gooding IOD require sufficient data streaks to accurately predict debris ephemerides and can struggle when arc lengths are too short (Zhang et al., 2024). Due to these short orbital arcs, this results in difficulties determining eccentricity. To avoid this issue, a major simplification is made assuming all eccentricities of detected objects are circular ( e = 0 ) (Abercromby et al., 2011). While MODEST does provide valuable insight into the GEO debris environment, uncertainty remains prevalent and should be accounted for, especially in the sub-10 cm debris regime and in-situ data-collection.

3.5.2.2 HUSIR

The push for a ground-based sub-10 cm detection facility began with the construction of the Haystack Radar in 1990 and the Haystack Auxiliary Radar (HAX) in 1994 (Murray et al., 2019). These highly sensitive radars were designed to detect small debris objects in LEO using radar cross-section (RCS) measurements. Haystack and HAX significantly improved the ability to observe debris throughout the millimeter and centimeter size regimes, providing crucial datasets for environmental models such as ORDEM and MASTER (Murray et al., 2019).

Both radars located in Tyngsborough, Massachusetts, utilize a beam park mode in which the radar beam is stationed at a fixed pointing geometry (Murray et al., 2019). Using this strategy, the radar systems are able to accurately characterize the orbital debris environment down to 3 cm (HAX) and 5 mm with Haystack at altitudes up to 1000 km (Murray et al., 2019). These detection capabilities fill critical observational gaps for LEO debris environmental assessments.

Despite these advancements, several limitations restrict these ground-based radar observation systems. The most prominent limitations Haystack and HAX face are the altitude restrictions for sub-10 cm debris. While the radar systems are capable of detecting sub-10 cm debris, this capability is largely confined to specific altitude regions within LEO. Additionally, while these efforts have filled gaps in LEO debris environmental models, gaps remain in models representing higher orbital regimes (GEO and MEO).

4 Research gaps and future work

The limitations described in previous sections highlight several research gaps within sub-10 cm debris models and model validation frameworks as shown in Table 3. In addition to these individual model limitations, direct cross-model comparisons reveal systematic disagreements in the sub-10 cm regime, as discussed in Section 3.4.

TABLE 3. Summary of model validation techniques including strengths, limitations, and future directions for sub-10 cm modeling.

The lack of observational and in-situ datasets with the sub-10 cm regime significantly limits the capabilities of flux-based and environmental models. As a result, there are no current datasets that contain the ability to capture the full spatial distribution of sub-10 cm debris across all orbital altitudes. Additionally, the reliance on legacy in-situ datasets, produced by space shuttle missions, introduces the second major research gap. While these datasets provided the foundational insight for the development of evolutionary and environmental models, the orbital environment is increasing exponentially every year, and updated datasets need to be created to combat this increase. The absence of sustained, spatially diverse, modern in-situ debris analysis missions limits the ability to accurately categorize the debris environment under current conditions. Finally, while the partial LEO orbital environment ( < 1000 k m ) has been evaluated through the LDEF, EURECA, HUSIR, and the modern data missions, advances in the characterization of the debris environment in higher LEO orbits, MEO, and GEO need to be made to provide safer debris predictions. Addressing these gaps requires improvements in both observational capabilities and modeling approaches.

One of the most important steps forward is the development of a modern in-situ measurement mission capable of measuring debris populations in numerous size regimes. Such a mission would be able to provide a modern characterization of the orbital debris environment from LEO to GEO while reducing the reliance of debris models on legacy datasets. Finally, space-based debris detection systems capable of detecting or measuring sub-10 cm debris would represent the most promising path forward for closing the current observability gaps. By operating within or in close proximity to the debris populations, these systems can overcome limitations associated with ground-based observations as described in (Kirk and Kilic, 2026). The implementation of such systems would significantly improve the safety of manned and unmanned space missions through the generation of high-fidelity sub-10 cm debris observations.

5 Conclusion

The modeling and characterization of the sub-10 cm orbital debris environment remains among the most prominent unresolved problems in debris environment prediction and spacecraft safety. Although many of these debris modeling frameworks share fragmentation equations, through the NASA Standard Breakup Model, each model differs in source population, validation datasets, and propagation assumptions.

As analyzed throughout this study, cross-model comparisons demonstrate the quantitative differences produced from these various assumptions. Model predictions from the current-version MASTER-8 and ORDEM 3.1 differ by a factor of two–three near the centimeter scale and diverge to an order of magnitude or more at the millimeter scale and lower. This divergence also extends into the total debris spatial distribution trends in the smaller debris regimes. These discrepancies can consequently impact spacecraft risk assessments, shielding requirements, and overall spacecraft safety.

International debris modeling efforts such as DAMAGE, SDEEM, and DELTA, as well as commercial observation networks and modern missions such as ADLER-1 and the SDS, demonstrate the growing global effort to improve the characterization of the debris environment. However, current debris identification datasets are limited by particle size, orbital region, and spatial coverage. As a result, the current available datasets remain insufficient in both quantity and spatial diversity to accurately constrain the sub-10 cm debris population in the near-Earth orbital regime.

Future improvements in debris environment characterization should prioritize missions that sample multiple altitude and inclination regimes rather than employing missions that remain confined to a single orbital regime for most of their operational lifetime. This can be done through the deployment of in-situ detection constellations similar to methodologies discussed within (Furumoto et al., 2017). Additionally, coupling in-situ constellation data collection mission with close-proximity optical detection methods, as discussed in Kirk and Kilic (2026), will help to provide broader validation data for debris prediction models. Ultimately, improving the characterization and validation of the sub-10 cm debris population is essential for maintaining the sustainability of near-Earth space for future utilization and exploration.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Generative AI was used for manuscript organization and grammatical editing. The authors take fully responsibility for the technical content, validation, and accuracy of data presented within the submitted manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Guglielmo S. Aglietti, The University of Auckland, New Zealand

Reviewed by: Gernot Groemer, Austrian Space Forum, Austria

Priya S. B. M, Sastra University, India